Re: [PATCH] tracing/uprobe: Fix double perf_event linking on multiprobe uprobe

[Masami Hiramatsu <mhiramat@kernel.org>]

On Mon, 20 Jan 2020 13:40:22 +0100
Peter Zijlstra <peterz@infradead.org> wrote:

> On Fri, Jan 10, 2020 at 10:45:39AM +0900, Masami Hiramatsu wrote:
> 
> > diff --git a/kernel/trace/trace_probe.h b/kernel/trace/trace_probe.h
> > index 4ee703728aec..03e4e180058d 100644
> > --- a/kernel/trace/trace_probe.h
> > +++ b/kernel/trace/trace_probe.h
> > @@ -230,6 +230,7 @@ struct trace_probe_event {
> >  	struct trace_event_call		call;
> >  	struct list_head 		files;
> >  	struct list_head		probes;
> > +	char				data[0];
> >  };
> 
> Would it make sense to make the above:
> 
> 	struct trace_uprobe_filter	filter[0];
> 
> instead? That would ensure that alignment is respected. While I think
> the current code works by accident.

Hmm, if we consider the alignment, shouldn't we allocate the structure
with the alignment gap? Currently it just added the 
sizeof(struct trace_uprobe_filter) when kzalloc().

In this case, I think we should introduce a new data structure,
trace_uprobe_event.

Thank you,

> 
> > @@ -264,6 +263,14 @@ process_fetch_insn(struct fetch_insn *code, struct pt_regs *regs, void *dest,
> >  }
> >  NOKPROBE_SYMBOL(process_fetch_insn)
> >  
> > +static struct trace_uprobe_filter *
> > +trace_uprobe_get_filter(struct trace_uprobe *tu)
> > +{
> > +	struct trace_probe_event *event = tu->tp.event;
> > +
> > +	return (struct trace_uprobe_filter *)&event->data[0];
> > +}
> 
> 


-- 
Masami Hiramatsu <mhiramat@kernel.org>